Depending on which font(s) you have installed, one or the other symbol in paragraph headings may not display correctly. To properly view the phonetic symbols on this page, you must have one of the following fonts installed on your system.

All are available freeware. I'm committed to keeping my recommendations to a) freeware fonts with b) decent looking IPA compliant symbols. If anybody has any recommendations for freeware Unicode fonts with good looking IPA support, I'll test them out and add them to the style sheet.

Solution for August 2003


"It could have been the ghost."

Glottal Stop, IPA 113,
[?], [ʔ]
The real evidence for the glottal stop here is the irregularity of the glottal pulses (voicing) at the beginning of the vowel. After tight closure, it takes a while for normal, regular vibration to get going, and the pulses start out like a sputtering engine.

Small Capital I + Subscript Tilde, IPA 319 + 406,
[I0], [ɪ̰]
Vowel, nice clear formants, good amplitude. So with vowels, we look first at the formants. The F1 (first, lowest, formant) is below the mid-range (500-600 Hz or so), so this vowel is moderately high. The second formant moves from the beginning of this vowel, between 125 and 150 msec, to the end, about 100 msec later. It starts quite high (mid-range for F2 is 1500-1600 Hz), up around 2000, and falls, slightly, to 1800 Hz or so by the end. So the high F2 tells us that this vowel is relatively front, quite front actually. The movement might be either 'real' movement in the vowel (i.e. evidence of a diphthong, or moving vowel of some kind), or it might be transitional, indicating a movement from (near) a vowel target to (near) a consonant target. Since there's a gap following, we can take at least some of this movement as transitional. So ignoring the transitional information (which will tell us about the place of the following consonant), we have a definitely front, fairly high vowel. This is English, so haul out your vowel charts and find a vowel in the front, mid-to-high range. I marked this vowel as creaky-voiced in part because of the creakiness from the preceding glottal stop, but also because some creakiness creaps in in the last few pulses as well. Hmm.

Lower-case T, IPA 103,
[t], [t]
A plosive is formed by a complete obstruction of the airway. The result is a gap in the spectrogram corresponding to the duration of the closure when there can be no resonance. So here we have one. Actually two, as there's a pretty abrupt transient suggesting a release at about 300 msec, but let's just concentrate on the first one. Since the next thing isn't resonant, but plosive, the transitions out of this first gap won't tell us anything about its place. Since its a gap, we don't have a lot of information during it to tell us anything, so we have to look to the transitions into the closure. As we mentioned before, it looks like the F2 is lowering, but not a whole lot. The lowest visible frequency that F2 ever gets to in the preceding vowel is about 1800 Hz. The F3 isn't doing a whole lot (actually, it seems to be lowering just a bit, but since that would lead us to the wrong answer I'm going to ignore it. X-) If the F2 is heading to a frequency just above the mid-F2 mark (1500-1600 or so), this is pretty typical of alveolars. However, since the F2 is lowering we might wonder if this isn't bilabial. It's probably not because even though the F3 might be lowering, the F1 and F4 are definitely not. Bilabial stricture can only lower formants, and that movement in F3 just isn't convincing, and F4 isn't cooperating at all. Velar is ruled out because, whatever the F3 is doing, it isn't "pinching" up with the F2, which would be more typical of velar transitions. The transient at about 300 msec is consisent with the alveolar conclusion, in that it is not strongest in the low frequencies or equally strong at all frequencies (more consistent with bilabial bursts), nor is it strongest in the F2-F3 range (more consistent with velars). So even if all signs don't point to alveolar, at least they don't obviously point anywhere else.

Lower-case K + Right Superscript H, IPA 109 + 404,
[kH], [kʰ]
The trick to spectrogram reading is both positive and negative reasoning. You hope that all the cues you can see will point to something in particular, or *not* point anywhere else. Everything should be 'consistent with' whatever hypothesis you're entertaining, whether it is 'evidence for', not evidence against, or ambivalent or equivocal. So the evidence here is that from 300 to about 375 msec there's another gap, so we're dealing with a plosive. There's no voicing bar, so it's voiceless. The transitions into the following vowel are slightly obscured due to the long period of voicelessness (i.e. aspiration) from the release of the stop to somewhere between 425-450 msec. That's not outrageously long for aspiration, but it's definitely well into the obviously aspirated category. So we know it's phonemically voiceless and probably initial in its syllable (as opposed to suffixed to the end of the preceding syllable). The transitions aren't amazingly helpful. The F1 doesn't seem to be doing much, The F2 seems to be dropping from just below 1500 Hz or so. F3, what you can see of it, seems to be stuck at about 2400 Hz or so, but rises a little once voicing starts. So the falling F2 doesn't suggest bilabial, and its starting frequency doesn't suggest alveolar. So we might entertain velar, but then we'd hope to see positive evidence, in the form of velar pinch. I'd be much happier of the F3 definitely started lower and rose into the vowel. But it doesn't. So while the transitions don't point toward velar, the definitely point away from bilbial or alveolar. So there's one more thing to consider, which is that burst. It's double There's a sharp transient at about 375 msec, and another 'twin' transient (actually a bit stronger, but the same 'shape' in terms fo frequency) about halfway between the first one and 400 msec. (I'd say about 387 msec, but you aren't supposed to make those kinds of scale judg(e)ments when you only have a scale in 100 msec intervals.) Double bursts are absolutely chararacteristic with velars (you see them occasionally with bilabials and laminals/dentals but they just don't look like this). And the bursts are definitely centered (strongest) in the F2 range, rather than lower (more consistent with bilabials) or higher (alveolar). So we have a couple of positive cuse (double burst and burst frequency) and a couple of negative cues (formants don't suggest anything clearly other than velar). So this is probably velar. And aspirated, as mentioned earlier.

Upsilon, IPA 321,
[U], [ʊ]
This is very reminiscent of schwa, but then think about where upsilon is on the typical English vowel chart. Okay, this F1 is about the same frequency as in the preceding vowel which we decided was just a little high. The F2 is below neutral, but nowhere near as low as a seriously round vowel might be, considering where the F1 is. The F3 is pretty neutral. So we've got a backish, highish vowel, that might be moving toward middish and centralish as it goes on. So starting in the higher-backer space and moving towards schwa. The movement toward schwa is pretty clear evidence of one of our English short/lax/non-peripheral vowels, and the only one that's high and back is transcribed with upsilon. Voilá.

Fish-hook R, IPA 124,
[R], [ɾ]
So at about 525 or so, we've got a tiny short little moment of radically decreased amplitude. It's almost a gap, but there's some resonance during, and it's just too short to be a decent plosive. So this is a brief 'interruption' to the sonority or resonance going on around it. Sounds like a flap to me. So this is phonemically a /t/ or /d/, but since this is phonetics is a flap/tap thing.

Lower-case H, IPA 146,
[h], [h]

So, this has formant structure, but no striations. It's all fuzzy and noisy. So this is a fricative. But with formants. What does that tell you?

Okay, I'll tell you. This is /h/. Voiceless source, either glottal or epiglottal friction, exciting all the open cavities of the vocal tract, just as voicing would. Review source-filter theory.

Schwa IPA 322,
[«], [ə]

Okay, so this is mid-to-high again, but just slightly more mid than the previous vowels. The F2 is dead smack in the neutral range. The F3 is a ittle low, but what do you want.

Lower-case V, IPA 129,
[v], [v]
Approximant V , ʋ IPA 150, is a possibility here, excpet that is usually thought of as sonorant and frictionless. This looks more like a plosive, except that there's some evidence (depending on your screen resolution) of friction here, between 700 and 750 msec, at least, which is what I take to be the duraiton of the segment, more or less. Hence fricative. And it looks weak, but striated at the bottom, so there you go. The transitions aren't particularly useful, but if anything the're trending downward, all of them, so this looks vaguely labial.

Lower-case B, IPA 102,
[b], [b]
So there's evidence of a plosive, in particular the burst. Now, having clued you in about double bursts, you will no doubt really, really want this to be a double burst. And perhaps it is, but it's not the same kind. The two bursty things don't look the same, and neither looks velar. They first one (and sort of the second one, is strongest in the low frequencies, rathe rthan the F2/F3 range, although that's arguable, I suppose. But they're both exceptionally weak.

Lower-case I, IPA 301,
[i], [i]
Well, we're back to a vowel. Mid-to-high again (someday I'll remember to have a variety of heights in a spectrogram), and mostly front. You do the math.

Lower-case N, IPA 116,
[n], [n]
Well, the firs thitng to notice is that it's definitely voiced, and resonant, and then that it's of greatly reduced amplitude from the surrounding vowel(s). So this is probably a sonorant consonant, probably a nasal, judging from the zeroes. If you now my voice, you recognize the frequency (near 1500 Hz) of the pole, which is indicative of my [n]. If you dont you have to find a way to convince yourself that the transitions are alveolar, but since there's another coronal coming up, your only real clue is the F3, which isn't amazingly helpful.

Eth, IPA 131,
[D], [ð]
So your first clue that there's something else beside the nasal going on ehre is the change in amplitude, discontinuity, whatever it is, at about 975 msec. Its not much, but it tells us that something changed there. What exactly that is is not really readable, except that the F3 and F4 transtions in the following vowel look decidedly alveolar. The F2 doesn't help. If you don't catch that, there's no real good clue that anything is going on (except the otherwise exceptional length of a nasal stop in a relatively weak position), and you just have to pick this up from context.

Schwa IPA 322,
[«], [ə]
Schwa, ah, schwa. Pretty classic, short vowel, quite low pitch of voice so probably stressless and reduced. Evenly spaced formants.

Lower-case G, IPA 110,
[g], [g]
I transcribed this as voiced, although now I'm not sure. There's a nivce velar-looking double burst, followed by evidence of velar transitions, so velar's the best guess here. Unaspirated, definitely, so phonemically /g/, at least.

Lower-case O + Upsilon, IPA 307 + 321,
[oU], [oʊ]
Okay, looking at the F1, it's higher than most of the previous vowels, which have mostly been mid to high. So this is clearly mid, through most of its duration, although there's something going on about 1400 msec and after we may want to pay attention to. The F2 starts around the neutral frequency, but drops rapidly, indicating increasing backness or rounding through this vowel. Hence this is a mid, backish vowel, getting backer, hence diphthong-y /o/. I don't know what's going on at 1400 msec. The pitch is rising, but I may just get a little creaky here. I can never tell when listening to my own voice. I bet there's something interesting going on here with the creakiness and the rising pitch. But whatever.

Lower-case S, IPA 132
[s], [s]
Well, you can't really ask for a better /s/, unless it didn't suddenly cut off in the low frequencies, which makes it look a little [S]/[s] (Esh)-like. So there you go.

Lower-case T + Right Superscript H, IPA 103 + 404,
[tH], [tʰ]
Well, except for the gap, this is more /s/. So the gap, given the position at the end of the syllable/word/utterance, is most likely /t/, since there's no evidence of labial or velar shaping. The real question is whether it's /sts/ or just /st/ with heavy release. Well, it turns out to be heavy release, but I'm not sure how you'd know that with nothing to compare it to.